WO2004071130A1 - 収音方法及び収音装置 - Google Patents
収音方法及び収音装置 Download PDFInfo
- Publication number
- WO2004071130A1 WO2004071130A1 PCT/JP2004/001261 JP2004001261W WO2004071130A1 WO 2004071130 A1 WO2004071130 A1 WO 2004071130A1 JP 2004001261 W JP2004001261 W JP 2004001261W WO 2004071130 A1 WO2004071130 A1 WO 2004071130A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- covariance matrix
- sound
- signal
- section
- sound source
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims description 54
- 239000011159 matrix material Substances 0.000 claims abstract description 247
- 230000005540 biological transmission Effects 0.000 claims abstract description 34
- 230000005236 sound signal Effects 0.000 claims abstract description 23
- 238000004364 calculation method Methods 0.000 claims description 62
- 238000001514 detection method Methods 0.000 claims description 44
- 238000001914 filtration Methods 0.000 claims description 23
- 230000035945 sensitivity Effects 0.000 claims description 19
- 238000006243 chemical reaction Methods 0.000 claims description 12
- 230000001131 transforming effect Effects 0.000 claims 2
- 238000012545 processing Methods 0.000 abstract description 11
- 230000001629 suppression Effects 0.000 description 27
- 238000010586 diagram Methods 0.000 description 26
- 230000007774 longterm Effects 0.000 description 11
- 230000006870 function Effects 0.000 description 9
- 230000003321 amplification Effects 0.000 description 8
- 238000003199 nucleic acid amplification method Methods 0.000 description 8
- 238000004088 simulation Methods 0.000 description 7
- 238000012935 Averaging Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 4
- 238000012937 correction Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000001228 spectrum Methods 0.000 description 4
- 230000002087 whitening effect Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 235000012149 noodles Nutrition 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M3/00—Automatic or semi-automatic exchanges
- H04M3/42—Systems providing special services or facilities to subscribers
- H04M3/56—Arrangements for connecting several subscribers to a common circuit, i.e. affording conference facilities
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0316—Speech enhancement, e.g. noise reduction or echo cancellation by changing the amplitude
- G10L21/0364—Speech enhancement, e.g. noise reduction or echo cancellation by changing the amplitude for improving intelligibility
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
- G10L2021/02161—Number of inputs available containing the signal or the noise to be suppressed
- G10L2021/02166—Microphone arrays; Beamforming
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/78—Detection of presence or absence of voice signals
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M2201/00—Electronic components, circuits, software, systems or apparatus used in telephone systems
- H04M2201/41—Electronic components, circuits, software, systems or apparatus used in telephone systems using speaker recognition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R3/00—Circuits for transducers, loudspeakers or microphones
- H04R3/005—Circuits for transducers, loudspeakers or microphones for combining the signals of two or more microphones
Definitions
- the present invention relates to a sound collecting method and a sound collecting device, and more particularly to a sound collecting method and a sound collecting device that collects transmission voices from a plurality of sound sources and adjusts and outputs respective volumes.
- the received signal level is greatly different.
- the volume of the reproduced sound varies greatly depending on the participant on the transmitting side, and it may be very difficult to hear.
- FIG. 17 is a block diagram showing a basic configuration of a conventional sound collecting apparatus disclosed in Japanese Patent Application Publication No. 8-250944, for example.
- This conventional sound collection device includes a microphone 41, a power calculation unit 42, an amplification factor setting unit 43, and an amplifier 44.
- the power calculation unit 42 obtains the long-term average power P ave of the signal received by the microphone 41.
- the long-term average power can be obtained by squaring the received signal and integrating over time.
- the amplification factor setting unit 43 calculates the long-term average power Pave of the received signal calculated by the power calculation unit 42 and the desired transmission level P set in advance.
- Set the amplification factor G based on pt .
- the amplification factor G is obtained, for example, by the following equation (1).
- the amplifier 44 amplifies and outputs the microphone reception signal at the set amplification factor G.
- the output signal power is at the desired transmission level P. pt and the volume is automatically adjusted.
- the amplification factor is determined based on the long-term average power, so that a delay of several seconds to several tens of seconds occurs before an appropriate amplification factor is set.
- the adaptation of the amplification factor setting is delayed every time the speaker switches, resulting in an inappropriate volume. The problem arises.
- An object of the present invention is to provide a sound collection device, a sound collection method, and a program for automatically adjusting the volume so that each speaker's voice has an appropriate volume even when the sound is collected by a microphone. Disclosure of the invention
- the sound collection method for collecting the sound emitted from the sound source by the microphones of multiple channels according to this light
- a state determination step including an utterance determination step of determining an utterance section from a sound reception signal received by the microphones of the plurality of channels;
- a sound collection device that collects sounds emitted from respective sound sources by microphones of multiple channels arranged in an acoustic space
- a state determination unit including an utterance determination unit that determines an utterance section from a sound reception signal received by each of the microphones of a plurality of channels;
- a sound source position detecting unit that detects a sound source position from the sound receiving signal when the utterance determining unit determines that the utterance period is an utterance section;
- a frequency domain conversion unit that converts the sound reception signal into a frequency domain signal;
- a covariance matrix calculation unit that calculates a covariance matrix of the sound reception signals of a plurality of channels in the frequency domain for each sound source;
- a covariance matrix storage unit that stores the covariance matrix for each sound source based on a detection result of the sound source position detection unit
- a filter coefficient calculation unit that calculates filter coefficients of a plurality of channels such that the transmission signal level for each sound source becomes a desired level, and converts the microphone reception signal into the plurality of channels. With the filter coefficient of, the filter of each channel is filtered,
- a sound collection method for collecting sound from at least one sound source by at least one microphone in an acoustic space in which a received signal is reproduced from a speaker
- the filter for at least one channel of the received sound signal so as to suppress the acoustic echo, which is the received signal component included in the received signal, based on the covariance matrix stored in the speech section and the received section.
- a evening coefficient calculating step for calculating the evening coefficient and the filter coefficient for the received signal,
- the sound collection device includes:
- At least one channel microphone for collecting transmission sound from a sound source and outputting a reception signal
- a state determination unit that determines an utterance section and a reception section from the reception signal and the reception signal; a frequency domain conversion unit that converts the reception signal and the reception signal into a frequency domain signal; A covariance matrix calculation unit that calculates a covariance matrix between the sound signal and the received signal in the frequency domain,
- a covariance matrix storage unit for storing the covariance matrix for each of the speech section and the reception section; and a sound reception signal of the at least one channel so as to suppress an acoustic echo of the reception signal based on the stored covariance matrix.
- a filter coefficient calculating unit for calculating the filter coefficient for the received signal and the filter coefficient for the received signal,
- a filter coefficient for the received signal and a filter coefficient for the received signal are set, and the received signal filter and the received signal filter for filtering and outputting the received signal and the received signal are provided;
- An adder that adds the output of the sound reception signal filter and the output of the reception signal filter, and outputs the addition result as a transmission signal
- the speaker can be appropriately controlled by using a plurality of microphones. Automatically adjusts the volume so that it is appropriate for each time.
- FIG. 1 is a block diagram showing a sound pickup apparatus according to a first embodiment of the present invention.
- FIG. 2 is a block diagram showing a configuration example of the state determination unit 14 in FIG.
- FIG. 3 is a block diagram showing a configuration example of a sound source position detection unit 15 in FIG.
- FIG. 4 is a block diagram showing a configuration example of a filter coefficient calculation unit 21 in FIG.
- FIG. 5 is a flowchart showing a first example of a sound collecting method using the sound collecting device of FIG.
- FIG. 6 is a flowchart showing a second example of the sound collection method using the sound collection device of FIG.
- FIG. 7 is a flowchart showing a third example of the sound collection method using the sound collection device of FIG.
- FIG. 8 is a block diagram showing a sound collection device according to a second embodiment of the present invention.
- FIG. 9 is a block diagram showing a configuration example of the state determination unit 14 in FIG.
- FIG. 10 is a block diagram showing a sound pickup device according to a third embodiment of the present invention.
- FIG. 11 is a block diagram showing a configuration example of a state determination unit 14 in FIG.
- FIG. 12 is a block diagram showing a sound collecting apparatus according to a fourth embodiment of the present invention.
- FIG. 13 is a block diagram showing a sound collecting device according to a fifth embodiment of the present invention.
- FIG. 14 is a block diagram showing a configuration example of a weighting factor setting unit 21H in FIG.
- FIG. 15 is a block diagram showing another configuration example of the weight coefficient setting unit 21 H in FIG.
- FIG. 16 is a block diagram showing a configuration example of the whitening section 21 J provided in FIG.
- FIG. 17 is a block diagram illustrating a configuration example of a covariance matrix storage unit 18 when a covariance matrix averaging function is provided in each embodiment.
- FIG. 18A is a diagram showing a speech waveform obtained by a simulation of speakers A and B before performing the processing according to the first embodiment.
- FIG. 18B is a diagram showing speech waveforms obtained by simulations of speakers A and B when the processing according to the first embodiment is performed.
- FIG. 19 is a diagram showing received and transmitted voice waveforms obtained by a simulation showing suppression of acoustic echo and noise according to the third embodiment.
- FIG. 20 is a block diagram showing a conventional sound collecting device. BEST MODE FOR CARRYING OUT THE INVENTION
- FIG. 1 is a block diagram of a sound collection device according to a first embodiment of the present invention.
- the sound collection device includes an M-channel microphone 11 i to 11 M arranged in an acoustic space, filters 12 to 12 M , an adder 13, and a state determination unit 14. And the sound source Position detector 15, frequency domain transformer 16, covariance matrix calculator 17, covariance matrix storage 18, sound pickup level estimator 19, filter coefficient calculator 2 1 It is composed of
- This embodiment detects the position of the sound source 9 ⁇ 9 ⁇ acoustic space, to keep asking the covariance matrix in the frequency domain of the received sound signals for each source position, filter using these Find the coefficient.
- the signal from each sound source is controlled to have a constant volume.
- each output sound signal of the microphones 11 1 to 11 ⁇ is not shown in the figure, but a digital signal obtained by converting the signal into a digital signal at a predetermined sampling frequency by a digital-to-analog converter. It is assumed that the received signal is The same applies to other embodiments.
- the state determination unit 14 detects an utterance section from a microphone sound reception signal received by the microphones 11 ⁇ to 11 ⁇ . For example, as shown in FIG. 2, in the state determination unit 14, all the reception signals from the microphones 11i to llM are added in an addition unit 14A, and the added output is a short-time average power calculation unit 14B.
- short-term average power for example, 0.1 to: average power between IS
- long-term average power for example, 1 to: average power between LOOs
- the sound source position detection unit 15 estimates the position of the sound source.
- a method of estimating a sound source position for example, there is a method based on a cross-correlation method.
- the number of microphones is assumed to be M (M is an integer of 2 or more), and the measured value of the delay time difference between the received signals obtained from the signals received by the i-th microphone 1 1 i and j-th microphone 1 lj is ij and I do.
- the measured value of the delay time difference between the received signals can be obtained from the cross-correlation between the signals and the maximum peak position.
- the measured value of the delay time difference between the received signals obtained from these positions is expressed by equation (2).
- f 3 ⁇ 4 -(" x i- x ) 2 + (Yi-Y-Yang + (Zi-Z) 2- ⁇ Xj- X) 2 + (yj -Y) 2 + (Zj-Z) 2 (2)
- c is the speed of sound.
- the measured value of the delay time difference between the received signals and the estimated value f y multiplied by the sound speed c and converted to the distance are the measured value and the estimated value ⁇ of the distance difference between the sound receiving positions, respectively.
- Equation (3) If a solution that minimizes the root-mean-square error e (q) in Equation (3) is obtained, the estimated sound source position that minimizes the error S between the measured value and the estimated value of the delay time difference between received signals can be obtained.
- equation (3) is a system of nonlinear equations that is difficult to solve analytically, so it is determined by numerical analysis using sequential correction.
- q q (4) where "is the step size of the correction and set the value of ⁇ > 0.
- q (u) represents q after u times of correction, and u
- q ( 0 ) (statement,, ⁇ .,.)
- 0 a predetermined initial value.
- grad represents a gradient and is expressed by the following equations (5) to (10). .
- FIG. 3 is a block diagram showing a functional configuration of the sound source position detection unit 15.
- the sound source position detection unit 15 includes a delay time difference measurement unit 15A, a multiplier 15B, a distance calculation unit 15C, a mean square error calculation unit 15D, a gradient calculation unit 15E, It comprises a comparison / determination section 15F and an estimated position update section 15G.
- the delay time difference measuring unit 15A during a period that is sound from one sound source 9 k
- y is measured with a delay time difference by the cross-correlation method based on the sound reception signals from the microphones 11 and 1 1 j.
- the multiplier 15B multiplies the measured delay time difference by the sound speed c to obtain a distance difference between the sound source and the microphone 111j.
- the distance calculation unit 15C calculates the distances ri and rj between the estimated sound source positions (, ⁇ , and the microphones 111, 1j) given from the estimated position update unit 15G using Equations (9) and (10).
- the estimated position updating unit 15G provides the distance calculating unit 15C with arbitrary initial values t 0 , f 0 , and Z 0 ) as the first estimated sound source positions.
- the root-mean-square error calculator 15D calculates the root-mean-square error e (q) by using equation (3) for all the pairs of (i, j) described above.
- the gradient calculator 15 F calculates the gradient grad e (q) of the mean square error e (q) by using equations (6), (7) and (8) using the current estimated sound source position and.
- the unit 15C calculates the updated ri, rj in the same manner as described above using the updated estimated position (sentence, ⁇ ,), and then, the mean square error unit 15D updates e (q), The gradient calculation unit 15E calculates the updated gradient grad e (q), and the comparison determination unit 15F determines whether the updated mean square error e (q) is smaller than the threshold value e th .
- the update of the estimated position ( ⁇ ) is repeatedly executed until all the elements of the gradient of the mean-square error grad e (q) become a sufficiently small value (smaller than e th ), and the position of the sound source 9 k (statement, ⁇ ) is estimated.
- the positions of other sound sources are similarly estimated.
- the frequency domain conversion unit 16 converts a sound reception signal of each microphone microphone into a frequency domain signal.
- the sampling frequency of the received signal is 16 kHz
- FFT fast Fourier transform
- the covariance matrix calculation unit 17 calculates the covariance of the microphone sound reception signal, and generates a covariance matrix. If the frequency domain transform signals of the microphone receiving signal obtained by the frequency domain transform unit 16 for each sound source 9 k are (0)) to 53 ⁇ 4 (), the MXM covariance matrix Rxx (co) of these signals Is generally represented by equation (1 1). f ⁇ 1 ( ⁇ ) ⁇ 1 ( ⁇ ) * ⁇ 1 ( ⁇ ) ⁇ 2 ( ⁇ ) * ⁇ 1 ( ⁇ ) ⁇ ) * ⁇ 2 ( ⁇ ) ⁇ 1 ( ⁇ ) * ⁇ 2 ( ⁇ ) ⁇ 2 ( ⁇ ) * ⁇ 2 ( ⁇ ) ⁇ ⁇ ( ⁇ ) *
- the covariance matrix storage unit 18 sets the covariance matrix based on the detection result of the sound source position detection unit 15.
- the variance matrix Rxx (c) is stored as the MXM covariance matrix R SkSk (co) of the received signal for each sound source 9 k .
- the sound collection level estimator 19 The sound pickup level P Sk of each sound source is calculated using the covariance matrix RskSk () of the received signal for each sound source 9 k stored in the covariance matrix storage unit 18 as follows:
- the element of the vector Ak may be a predetermined value aki, a k2 , ..., a k M.
- elements of the weighting mixing base vector A k for each sound source 9 k has a significant value as elements corresponding to the near have the microphone to the sound source 9 k.
- aki a k2 , ..., a k M.
- H represents the complex conjugate transpose
- a k () HR sksk (w) A k (co) can be expanded as follows.
- Equation (1 2) converts the value of the power spectrum sample represented by ⁇ ( ⁇ ) represented by the equation (13) into the band 0 to W (the number of samples) of the frequency domain signal generated by the frequency domain transform unit 16. It means that the average power P Sk of the received sound signal is obtained by adding over and dividing by W.
- the value of the weighting coefficient a kl is determined so as to give the maximum weight to the sound reception signal of the microphone 11 (first channel). Therefore , the weighting factors a k2 , a k3 , and akM for the sound reception signals of the other channels are determined to be smaller than a kl .
- the weighting coefficient of the weighting mixing vector A k (6)) for each sound source 9 k is set to be such that, for example, the S / N of the output audio signal corresponding to the sound source 9 k is large and the room reverberation is small.
- the optimum value is determined experimentally in advance according to the directivity and arrangement of the sound source and the arrangement of the sound source.
- the filter coefficient calculation unit 21 calculates a filter coefficient for collecting sound emitted from each sound source at a desired volume.
- the filter coefficients of 1 2 i to l 2 M connected to each microphone are converted to the frequency domain! ! ) ⁇ !!! ⁇ ).
- ⁇ ( ⁇ ) be a matrix of these filter coefficients by the following equation (14).
- ⁇ ⁇ ( ⁇ ) a frequency domain converted signal received sound signal at each microphone period k-th sound source 9 k is pronounced Xsk, i ⁇ Xsk, and M.
- the condition required for the filter coefficient matrix ⁇ ( ⁇ ) is that the microphone receiving sound signal is filtered by the filter coefficient matrix ⁇ ( ⁇ ), and the signal after each filtering is added when the filtered signal is added.
- the level of the component is the desired level P. pt . Therefore, a signal obtained by filtering and adding the signal of each sound source 9 k is a signal obtained by multiplying the weighted mixing vector A k ( ⁇ ) for the sound reception signals of the microphones 11 i to l 1 M by the desired gain.
- the following equation (15) is an ideal condition.
- c sk is a weighting factor that gives a sensitivity constraint on the th sound source position.
- Sensitivity constraint means flattening the frequency characteristics of the sound pickup device with respect to the sound source position. As this value increases, the sensitivity constraint on the sound source becomes stronger, and sound collection with flatter frequency characteristics becomes possible, but the deterioration of the frequency characteristics with respect to other sound source positions increases. Therefore, it is usually preferable to set C Sk to a value of about 0.1 to 10 and to constrain the sensitivity of all sound sources with good balance.
- FIG. 4 shows a functional configuration block diagram of the filter coefficient calculation unit 21 for calculating the filter coefficient represented by the equation (16).
- the covariance matrices R S1S1 to R SKSK corresponding to the respective sound sources 9 i to 9 ⁇ given from the covariance matrix storage unit 18 are weighted by the multipliers 21A1 to 21AK in the weight coefficient setting unit 2 respectively. Multiplied by the weighting factors C S1 to C SK set by 1H.
- the sound pickup levels P S1 to PSK for the sound sources 9 ⁇ to 9 ⁇ estimated by the sound pickup level estimating section 19 are desired output levels P predetermined by the square ratio calculating sections 21B1 to 21BK. square ratio of the pt (P.
- the filter coefficient obtained by the fill evening coefficient calculator 2 1 (0)), 11 2 (0), ..., ⁇ ⁇ ( ⁇ ) is Filter 1 2 i-l 2 M, Filters the received signals from the microphones 1 1 ⁇ to 11 M.
- the signal after the filter processing is added by the adder 13 and output as an output signal.
- step S2 the state determination unit 14 periodically performs utterance detection, and When a talk is detected, the sound source position is detected by the sound source position detection unit 15 in step S3.
- step S4 it is determined whether the detected sound source position matches any of the previously detected sound source positions. If there is a match, in step S5 the covariance matrix R xx corresponding to the sound source position is determined. (c) is newly calculated by the covariance matrix calculation unit 17, and the covariance matrix of the corresponding area in the covariance matrix storage unit 18 is updated in step S 6.
- step S4 If it does not match the previously detected sound source position in step S4, K is increased by 1 in step S7, and in step S8 the covariance matrix R xx (co) corresponding to that sound source position is shared.
- the variance matrix calculation unit 17 newly calculates, and the covariance matrix is stored in a new area of the covariance matrix storage unit 18 in step S9.
- the sound collection level estimation unit 19 estimates the sound collection level from the covariance matrix stored in step S 10, and in step S 11 filters using the estimated sound collection level and the covariance matrix.
- the coefficient calculation unit 21 calculates the filter coefficient ( ⁇ ) to ⁇ ⁇ ( ⁇ ), and in step S12, updates the set filter coefficient of the filter 12 1 to 12 ⁇ .
- step S1 the maximum value of the number of sound sources is set to K fflax in advance, and the initial value of the number of sound sources K is set to 0.
- steps S2 to S6 utterance detection is performed in the same manner as in FIG. 5, and when utterance is detected, the sound source position is detected, and it is determined whether or not the detected sound source position matches one detected in the past. If there is a match, calculate the covariance matrix and save it in the corresponding area.
- step S 4 If the detected sound source position does not match the past detection sound source position in step S 4, 1 is added to the K in Sutetsu flop S 7, K is you determine becomes larger than the maximum value K X in step S 8. If it does not exceed the maximum value Kinax , a covariance matrix for the detected position is calculated in step S9 , and the covariance matrix is stored in a new area in step S10 .
- K K Iiax is set in step S11, and in step S12 , the covariance matrix stored in the covariance matrix storage unit 18 The most recently updated covariance matrix is deleted, and the new covariance matrix calculated by the covariance matrix calculation unit 17 in step S13 is stored in that area in step S14.
- the estimated pickup level of each sound source is obtained from the covariance matrix, and the filter Calculate coefficients and set filter 1 2 1 2 M.
- the maximum value of the number of sound sources K is limited to K MX by There is an advantage that the storage area of the covariance matrix storage unit 18 can be reduced as compared with the method.
- step S1 the initial value of the number of sound sources K is set to 0, and in step S2, the state detection unit 14 periodically performs an utterance detection operation, and when the utterance state is detected, in step S3, the sound source position detection unit 1 ⁇ Performs sound source position detection in step 5.
- step S4 it is determined whether the detected sound source position matches any of the detected sound source positions. If there is a match, the process returns to step S2 without performing the update process.
- step 5 K is increased by one, and in step S6, a covariance matrix R SkSk (w) corresponding to the sound source is newly calculated in the covariance matrix calculation unit 17 and in step S7, the covariance matrix storage unit 1 8 with the corresponding saving in the new area MA k of estimated sound pickup level by sound pickup level estimating unit 1 9 with the covariance matrix in step S 8, all of the covariance matrix in step S 9
- the updated filter coefficient is calculated by the filter coefficient calculator 21 using the estimated sound pickup level and the filter coefficient calculated by the filter coefficient calculator 21.In step S10, the updated filter coefficients are updated to the filters 12 to 12M. Set and return to step S2.
- the sound source position is estimated from the sound reception signals of a plurality of microphones, the covariance matrix of the sound reception signal is obtained for each sound source, and the filter coefficient for adjusting the volume for each sound source position is determined. Then, the microphone reception signal is filtered with those filter coefficients to obtain an output signal whose volume is adjusted for each speaker position.
- the sound source position detection unit 15 estimates the coordinate position of each sound source 9 k has been described. However, the sound source direction, that is, each sound source 9 k for the array of microphones 11 i to l 1 M is described. May be obtained.
- FIG. 8 is a functional block diagram of a sound collection device according to a second embodiment of the present invention.
- the sound pickup device of this embodiment includes a microphone 11 to 11 M, a filter 12 to 12 M, an adder 13, a state determination unit 14, a sound source position detection unit 15, and a frequency domain. It comprises a conversion unit 16, a covariance matrix calculation unit 17, a covariance matrix storage unit 18, a sound collection level estimation unit 19, and a filter coefficient calculation unit 21.
- noise suppression is added to the sound collection level adjustment of the sound collection device according to the first embodiment of the present invention.
- the state determination unit 14 detects an utterance section and a noise section from the power of the microphone reception signal received by the microphones 11i to 11M. As shown in FIG. 9, the state determination unit 14 has a configuration in which a noise determination unit 14F is further added to the configuration of the state determination unit 14 in FIG. For example, as in the first embodiment, the short-time average power P avS and the long-term average power P avL are calculated for each microphone reception signal in a short-term average power calculation section 14 B and a long-term average power calculation section.
- the ratio R p P avS / P avL between the short-term average power and the long-term average power is determined by the division unit 14 D, and is compared with the utterance threshold Pthu by the utterance determination unit 14 E. If it exceeds, it is determined to be an utterance section.
- Noise determination unit 1 4 F determines the power ratio R p as compared to noise threshold P noodles, a noise segment if smaller than that.
- the sound source position detection unit 15 detects the sound source position in the same manner as in the first embodiment of the present invention.
- frequency domain transform section 1 6 covariance matrix is converted into a frequency domain signal received sound signal from each sound source 9 k pronunciation section and the noise section their respective microphones 1 1! To 1 1 M This is given to the operation unit 17.
- the covariance matrix calculation unit 17 calculates the covariance matrix R SkSk (co) of the frequency domain received signal for each sound source 9 k in the same manner as in the first embodiment of the present invention. Further, a covariance matrix R NN (c) of the frequency domain received signal in the noise section is calculated.
- the covariance matrix R sksk (c) and the noise interval for the sound source 9 1; 9 ⁇ are shared based on the detection result of the sound source position detection unit 15 and the determination result of the state determination unit 15. Save the variance matrix R NN (c) in the regions MAi, ..., ⁇ , ⁇ + ⁇ .
- the sound pickup level estimating section 19 calculates the sound pickup level P sk of each sound source as in the first embodiment of the present invention. Estimate similarly.
- the filter coefficient calculating section 21 collects sound emitted from each sound source 9 k with a desired volume and calculates a filter coefficient for suppressing noise.
- the conditions for noise suppression are determined.
- the frequency domain converted signal of the microphone received sound signals in a noise period ⁇ ⁇
- Microphone received sound signals in a noise period ⁇ ⁇ > 1 ( ⁇ ) ⁇ ⁇ , ⁇ ( ⁇ ) is if the zero filter 1 2 i-l 2 M and adder 1 3 after passing through, and can suppress noise Therefore, the conditional expression for noise suppression is expressed by Expression (17).
- C N is a constant of the weight for the amount of noise cancellation. As the value increases, the amount of noise cancellation increases. However, increasing C N weakens the sensitivity constraint on the sound source position and greatly degrades the frequency characteristics of the picked up signal. Therefore, C N is usually set to an appropriate value of about 0.1 to 10.0. The meanings of the other symbols are the same as in the first embodiment.
- the filter coefficient obtained by the equation (18) is set to a filter 12i to l2M, and the microphone reception signal is respectively filtered.
- the filtered signal is added by the adder 13 and output as an output signal.
- the second embodiment of the present invention can perform noise suppression in addition to the sound collection level adjustment effect of the first embodiment of the present invention.
- FIG. 10 is a functional block diagram of a sound pickup device according to a third embodiment of the present invention.
- the sound pickup device of this embodiment includes a speaker 22, a microphone 11 t to 1 M, a filter 12 ⁇ to 12 M , 23, an adder 13, and a state determination unit 14.
- a speaker 22 for reproducing a received signal and a filter 23 for filtering the received signal are added to the sound pickup apparatus according to the second embodiment of the present invention shown in FIG.
- a sound echo which is a speaker reproduction signal component picked up by the microphones 11i to 11M, is suppressed.
- the state determination unit 14 further includes a short-time average power P ′ avS and a long-time average power P avL of the received signal, which are added to the configuration of the state determination unit 14 in FIG.
- Average power calculator 14 B ′, long-term average power calculator 14 C ′, and their ratio R ′ p P ′ avS / P ′
- Divider 14 D ′ for calculating avL and their ratio R ′ p is compared with a predetermined threshold value R thR of the received signal, and if it is larger than the threshold value, the received signal is determined to be in the receiving section 14 G, the utterance determining unit 14 E, the noise determining unit 14 F, a state determination unit 14H that determines a state based on the determination result of the reception determination unit 14G is added.
- the state determination unit 14H determines whether the reception period is irrespective of the determination results of the other speech determination units 14E and the noise determination unit 14F.
- the utterance section or the noise section is determined in accordance with the determinations of the utterance determination section 14E and the noise determination section 14F as in FIG. Is determined.
- the sound source position detection unit 15 detects the sound source position in the same manner as in the first embodiment of the present invention.
- the covariance matrix ⁇ ( ⁇ ) of the frequency domain transform signal ⁇ ⁇ ⁇ ⁇ ( ⁇ ) of the microphone reception signal and the frequency domain transform signal ⁇ ( ⁇ ) of the reception signal is calculated by the following equation (19).
- * represents a complex conjugate
- the covariance matrix storage unit 18 stores a covariance matrix Rxx (o) based on the detection result of the sound source position detection unit 15 and the determination result of the state determination unit 14 for each sound source 9 k in the utterance interval.
- the sound pickup level estimating unit 19 calculates the sound pickup level Psk of each sound source 9 k using a predetermined weighted mixing vector composed of the covariance matrices Rsisi and RSKSK for each sound source, and M + 1 elements for each sound source.
- ⁇ ⁇ ⁇ ( ⁇ ), ⁇ , ⁇ ⁇ ⁇ ⁇ ( ⁇ ) is obtained by the following equation (20).
- the filter coefficient calculation unit 21 calculates a filter coefficient for collecting sound emitted from each sound source at a desired volume.
- the filter coefficients of the filters 12 to 12 M connected to each microphone are converted to the frequency domain! ! ) ⁇ !!! ⁇ ) and F (co) is obtained by converting the filter coefficient of the filter 23 for filtering the received signal into the frequency domain.
- ⁇ ⁇ ⁇ ⁇ ( ⁇ ) be a matrix of these filter coefficients according to equation (21).
- the conditions required for the filter coefficient matrix ⁇ ( ⁇ ) are as follows: the microphone reception signal and the transmission signal are filtered using the filter coefficient matrix ⁇ ( ⁇ ), respectively, and the filtered signal is added. Sometimes, the acoustic echo signal and the noise signal are eliminated, and only the transmitted voice signal is transmitted at the desired level.
- CE is a weight constant for the amount of acoustic echo cancellation, and the larger the value, the greater the amount of acoustic echo cancellation.
- CE is usually set to an appropriate value of about 0.1 to 10.0.
- the meanings of the other symbols are the same as in the second embodiment.
- the volume is adjusted and the filter coefficients are determined so as to suppress noise. be able to.
- the filter coefficients are set in the filter 1 2 i-l 2 M, 2 3, to fill evening rings each microphone received sound signal and the received signal.
- the filtered signal is added by the adder 13 and output as an output signal.
- the other parts are the same as those of the second embodiment of the present invention, and the description is omitted.
- the third embodiment of the present invention can perform acoustic echo suppression in addition to the sound collection level adjustment and noise suppression effects of the second embodiment of the present invention.
- the case where the acoustic echo suppression function is added in the second embodiment is shown.
- the acoustic echo suppression function may be further added to the first embodiment.
- the noise determination unit 14 F is removed in FIG. 11 showing the details of the state determination unit 14 in FIG. 10, and the covariance matrix calculation unit 17 in FIG. NN (co) is not calculated. Therefore, the calculation of the filter coefficient in the filter coefficient calculator 21 is represented by the following equation.
- the third embodiment of FIG. 10 shows an embodiment in which an acoustic echo canceling function is further added to the sound collection level adjusting function and the noise suppressing function of the second embodiment, but has only the noise suppressing and acoustic echo canceling functions.
- You may comprise as a sound collection device.
- Figure 12 shows an example of the configuration. As shown in FIG. 12, the configuration of this embodiment removes the sound source position detection unit 15 and the sound collection level estimation unit 19 from the configuration of FIG.
- the covariance matrix Rss ()), the covariance matrix of the received signal ⁇ ⁇ ( ⁇ ), and the covariance matrix of the noise signal ⁇ , ⁇ ( ⁇ ) are calculated, and the storage area MAs, MA E, and is configured to store each of the MA N. Also, in order to realize this acoustic echo cancellation function, at least one microphone is required. Here, an example using M microphones is shown.
- the state determination unit 14 determines an utterance section, a reception section, and a noise section from the reception signal and the reception signal received by the microphones 12i to l2M, similarly to the embodiment of FIG. The specific configuration and operation are the same as those shown in FIG.
- the received signal and the received signal are converted into a frequency-domain received signal ⁇ ( ⁇ ) ⁇ : ⁇ ⁇ ( ⁇ ) and a frequency-domain received signal ⁇ ( ⁇ ) by the frequency domain conversion unit 16 and the covariance matrix is calculated. Part 17 is given.
- the covariance matrix calculation unit 17 generates a covariance matrix of the frequency domain received signal and the received signal.
- the covariance matrix Rxx (c) of the frequency domain transformed signal of the microphone received signal ⁇ ⁇ ⁇ )): ⁇ ⁇ ( ⁇ ) and the frequency domain transformed signal of the received signal ⁇ ( ⁇ ) is given by the following equation (27). It is calculated from:
- the covariance matrix storage unit 18 calculates the covariance matrix Rxx (co) based on the determination result of the state determination unit 14 with the covariance matrix Rss (a) of the received signal and the received signal in the receiving section.
- the covariance matrix RNN () of the received signal and the received signal in the noise section and the covariance matrix REE (O) of the received signal and the received signal in the utterance section are stored.
- the filter coefficient calculation unit 21 collects the transmission sound emitted from the sound source, and calculates a filter coefficient for suppressing acoustic echo and noise.
- the filter coefficients of the filters 12 i to l 2 M connected to the microphones 1 1 j to l 1 M are converted to the frequency domain as ⁇ ⁇ ) ⁇ : ⁇ ⁇ ( ⁇ ).
- F (co) is obtained by converting the filter coefficient of the filter 23 to be filtered into the frequency domain.
- ⁇ ( ⁇ ) denote the evening coefficient as a matrix according to equation (28).
- Xs, i (co) to Xs, M (co) be the frequency domain converted signal of the microphone received signal in the speech section
- 3 ⁇ 4 ( ⁇ ) be the frequency domain converted signal of the received signal.
- the condition required for the filter coefficient matrix ⁇ ( ⁇ ) is that the microphone reception signal and the transmission signal are respectively filtered by the filter coefficient matrix ⁇ ( ⁇ ), and the signals after the filtering are added. In addition, the acoustic echo signal and the noise signal are eliminated, and only the transmitted voice signal is transmitted.
- CE is a weight constant for the amount of acoustic echo cancellation, and the larger the value, the greater the amount of acoustic echo cancellation.
- CE is usually set to an appropriate value of about 0.1 to 10.0. The meanings of the other symbols are the same as in the second embodiment.
- the filter coefficient can be determined so as to adjust the volume and suppress noise.
- the filter coefficients obtained by the equation (32) are set to the filters 12 to 12M and 23, and the microphone reception signal and the reception signal are respectively filtered.
- the filtered signal is added by the adder 13 and output as an output signal.
- the other parts are the same as those of the second embodiment of the present invention, and the description is omitted.
- the fourth embodiment of the present invention can perform acoustic echo suppression in addition to the noise suppression effect.
- FIG. 13 shows a fifth embodiment.
- the fifth embodiment differs from the fourth embodiment shown in FIG. 12 in that the sound source position is detected in the utterance section, the covariance matrix is obtained for each sound source position, and these are stored. Find the matrix and save it. Next, using these stored covariance matrices, filter coefficients for suppressing noise and acoustic echo are determined. By filtering the microphone reception signal and the reception signal with these filter coefficients, a transmission signal in which noise and acoustic echo are suppressed is obtained.
- the configuration of the fifth embodiment is the same as the configuration of the third embodiment shown in FIG. 10 except that the sound pickup level estimation unit 19 is removed.
- the state determination unit 14 detects an utterance section, a reception section, and a noise section as in the third embodiment.
- the sound source position detection unit 15 estimates the position of each sound source 9k .
- the method of estimating the sound source position is the same as that of the first embodiment shown in FIG.
- the received sound signal and the received speech signal are converted into a frequency domain signal by the frequency domain conversion section 16 and provided to the covariance matrix calculation section 17.
- the covariance matrix calculation unit 17 calculates the covariance matrix Esi S1 (w E. maraud ( ⁇ ) of the received signal and the received signal for each sound source 9 k , the covariance matrix ⁇ ⁇ ( ⁇ ) of the receiving section, calculating a covariance matrix E Nyunyu the noise section (omega).
- covariance matrix storage unit 1 8 based on the position detection result of the determination result of the state determination unit 1 4 and a sound source position ⁇ detecting section 1 5, the covariance matrix H S1S1 (0)) ⁇ : R SKSK (o)), E (co), and R NN (c) are stored in the corresponding areas ⁇ to ⁇ , ⁇ ⁇ + 1 , and MA K +2.
- the filter coefficient calculating section 21 calculates a filter coefficient for suppressing acoustic echo and noise when the transmitted voice is collected. As in the third embodiment, if the conditional expression for the filter coefficient matrix ⁇ ( ⁇ ) is solved by a least squares solution, (33)
- C S1 to C SK are the weight constants of the sensitivity constraint for each sound source
- C E is the weight constant for the acoustic echo cancellation amount
- C N is the weight constant for the noise cancellation amount.
- fill evening coefficients obtained by is set to fill the evening 1 2 1 ⁇ 1 2 3 ⁇ 4 1, 2 3, filters the microphone received sound signal and the received signal, respectively.
- the signal after the filtering is added by the adder 13 and output as an output signal.
- the other parts are the same as those of the third embodiment of the present invention, and the description is omitted.
- the sensitivity can be constrained for a plurality of sound source positions, and it is possible to maintain the sensitivity even for sound sources spoken in the past. Therefore, even if the position of the sound source is moved, the sensitivity is maintained for the sound source that was uttered in the past, so that there is an advantage that the sound quality of the head of the transmitted voice does not deteriorate.
- a sound collecting device according to a sixth embodiment of the present invention will be described.
- the weighting coefficients C S1 to C SK of the sensitivity constraint for each sound source position 9 k of the sound pickup apparatuses according to the first to third and fifth embodiments of the present invention are described. Is changed over time.
- the weighting coefficients C S1 to C SK of the time-varying sensitivity constraints for the sound sources 9 i to 9 ⁇ are set smaller for sound sources spoken in the past.
- the weighting factor Csk becomes smaller as the elapsed time from the detection time of each sound source position already detected to the detection time of the most recently detected sound source position becomes longer.
- the weighting coefficient Csk is reduced in the order of detection time of the K sound source positions.
- FIG. 14 is a functional configuration block diagram of the weighting factor setting unit 21H that performs the first method.
- the clock 21H1 weighting factor setting unit 2 1 H is for outputting a time, time storage unit to overwrite the detection time time t the number k representing the sound source 9 k detected for each detection source position as an address 21H2
- a weight coefficient determination unit 21H3 is the sound source position detection time recorded in the time recording unit 21H2.
- a predetermined value C s is given as a weighting factor C Sk to the sound source of the number k (t) detected this time based on the detection time t k for each of the other numbers k ⁇ k (t).
- the sensitivity constraint weighting factors C S1 to C SK are determined for each sound source, and given to multipliers 21A1 to 21AK.
- FIG. 15 is a functional configuration block diagram of the weighting factor setting unit 21H for implementing the second method.
- This example includes a clock 21H1, a time recording unit 21H2, an order determination unit 21H, and a weight coefficient determination unit 21H5.
- the weight coefficient determination unit 21H5 gives a predetermined value C s to the most recent detected sound source 9 k (1) as a weight coefficient C Sk (1) .
- These C Sk (i) to C sk (K) are rearranged in the order of ⁇ k (l),..., K (K) ⁇ and output as weighting factors C S1 ,..., C SK .
- the value of q is a predetermined value of 0 ⁇ q ⁇ l.
- the sensitivity constraint on the sound source position that has been uttered in the past can be reduced.
- the number of sound sources whose sensitivity is restricted can be reduced as compared with the sound pickup apparatuses according to the first to third embodiments of the present invention, and the sound pickup level adjustment performance and the noise and acoustic echo suppression performance can be improved. Can be.
- a sound pickup device according to a seventh embodiment of the present invention will be described.
- the sound pickup apparatus is configured to whiten the covariance matrix ⁇ ( ⁇ ) in the filter coefficient calculation unit 21 of the sound pickup apparatuses according to the first to sixth embodiments of the present invention. It is characterized.
- FIG. 16 shows a functional configuration of one of the whitening units 21J1 to 21JK indicated by a broken line in the filter coefficient calculation unit 21 of FIG.
- the whitening unit 21J is composed of a diagonal matrix calculation unit 21JA, a weighting unit 21JB, an inverse operation unit 21JC, and a multiplication unit 21JD. You.
- the diagonal matrix calculation unit 21JA generates a diagonal matrix diag (3 ⁇ 4 «( ⁇ )) of the given covariance matrix R xx (co).
- the weighting unit 21JB gives the weight to the diagonal matrix by calculating the following equation using the weight based on the predetermined matrix D of M rows or M + 1 rows.
- the filter coefficient obtained in the filter coefficient calculator 21 does not change with respect to changes in the spectrum of the transmission signal, the reception signal, and the noise signal. As a result, there is no change in the sound pickup level adjustment performance and the acoustic echo and noise suppression performance due to the spectrum change, and steady sound pickup level adjustment, acoustic echo, and noise suppression can be obtained.
- a sound collecting device according to an eighth embodiment of the present invention will be described.
- the sound collection device includes: a covariance matrix storage unit 18 of the sound collection device according to the first to seventh embodiments of the present invention;
- the covariance matrix newly calculated by the covariance matrix calculation unit 17 is averaged and stored as a current covariance matrix.
- the averaging of the covariance matrix is performed, for example, by the following method.
- the already stored covariance matrix is Rxx ,. ld (c), and the covariance matrix newly calculated by the covariance matrix calculation unit 17 is 3 ⁇ 4X, new (CD).
- FIG. 17 shows a functional configuration of the covariance matrix storage unit 18 and the averaging unit 18A provided therein.
- the averaging unit 18A includes a multiplier 18A1, an adder 18A2, and a multiplier 18A3.
- the covariance matrix RskSk (o) corresponding to the sound source 9 k calculated by the covariance matrix calculation unit 17 is given to the multiplier 18A1 as a new covariance matrix R SkSk , new (co), and (1-p)
- the multiplication is performed, and the multiplication result is provided to the adder 18A2.
- the covariance matrix Rsksk (o) corresponding to the sound source 9 k is read from the storage area 18B , and the old covariance matrix R SkSk,. It is given to the multiplier 18A3 as i d (co) and is multiplied by the constant P. As a result of the multiplication output of the multiplier 18A1 by the adder 18A2 (1- p) R sksk, summed with new new (), the resulting mean covariance matrix Rsksk (co) is overwritten in the storage area corresponding to the sound source 9 k Is done.
- the present invention can be realized by dedicated hardware, or a program for realizing the function is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read by a computer. Can be executed.
- the computer-readable recording medium is a storage device such as a floppy disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a nonvolatile semiconductor memory, an internal or external hard disk, and the like.
- a computer-readable recording medium is one that dynamically holds a program for a short time (transmission medium or transmission), such as a case where a program is transmitted via the Internet. This includes programs that hold programs for a certain period of time, such as volatile memory inside a computer system used as a server.
- FIG. Figure 8B shows the simulation results when microphones were placed at each corner of a 20 cm square.
- the simulation conditions were as follows: Number of microphones: 4, Signal-to-noise ratio: Approx. 20 dB, room reverberation time: 300 ms, Number of speakers: 2 (Speaker A is 50 cm perpendicular to one side of the square from the center of the four microphone microphones) Speaker B is 200 cm from speaker A at 90 ° from speaker center.
- Figure 18A shows the microphone reception signal waveform when speaker A and speaker B sound alternately under these conditions.
- FIG. 18B shows a waveform after processing according to the present invention.
- the amplitudes of the speech waveforms of speaker A and speaker B are almost the same, confirming the effect of the sound collection level adjustment.
- FIG. 19 shows a simulation result according to the third embodiment shown in FIG.
- the simulation conditions were as follows: the number of micros M: 4, the transmission signal to noise ratio before processing: 20 dB, the transmission signal to acoustic echo signal ratio: -10 dB, and the room reverberation time: 300 msec. Under these conditions, transmission and reception are repeated alternately, and the transmission signal level at that time is shown in Figure 19.
- Line A is the transmission signal level before processing
- line B is the transmission signal level after processing according to the third embodiment. From the above results, it can be seen that the third embodiment reduces the acoustic echo signal by about 40 dB and the noise signal by about 15 dB, confirming that the embodiment of the present invention is effective.
- a sound source position is detected from signals received by a plurality of microphones, and a filter coefficient is calculated based on a covariance matrix of a speech section for each sound source position. Then, by filtering and adding the microphone reception signals with these filter coefficients, a transmission signal whose volume is adjusted for each sound source position can be obtained.
- the filter coefficient is determined using the covariance matrix of the noise section in addition to the covariance matrix of the speech section in the first embodiment, so that the sound pickup level can be adjusted. Noise suppression becomes possible.
- the filter coefficient is determined using the covariance matrix in the reception section in addition to the covariance matrix in the utterance section.
- the received signal is reproduced by the speaker, and the collected acoustic echo is suppressed by determining the filter coefficient using the covariance matrix of the utterance section and the covariance matrix of the received section. can do.
- the filter coefficient is determined using the covariance matrix of the noise section in addition to the covariance matrix of the speech section and the reception section, thereby further suppressing noise. Becomes possible.
- the weight of the smaller sensitivity constraint is applied to the covariance matrix of the weaker pronunciation when calculating the filter coefficients.
- the covariance matrix is whitened, so that the sound pickup level with little influence due to the change in the signal spectrum is obtained. It enables adjustment, noise suppression, and acoustic echo suppression.
- the weighted average covariance matrix with the covariance matrix already stored in the corresponding area is used.
- a more accurate covariance matrix can be obtained, and a filter coefficient with improved performance of sound collection level adjustment, noise suppression, and acoustic echo suppression can be determined.
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Circuit For Audible Band Transducer (AREA)
- Obtaining Desirable Characteristics In Audible-Bandwidth Transducers (AREA)
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
DE602004006967T DE602004006967T2 (de) | 2003-02-07 | 2004-02-06 | Telefonkonferenzverfahren und telefonkonferenzsystem |
EP04708849A EP1592282B1 (en) | 2003-02-07 | 2004-02-06 | Teleconferencing method and system |
US10/510,955 US7716044B2 (en) | 2003-02-07 | 2004-02-06 | Sound collecting method and sound collecting device |
JP2005504891A JP4104626B2 (ja) | 2003-02-07 | 2004-02-06 | 収音方法及び収音装置 |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2003030676 | 2003-02-07 | ||
JP2003-030676 | 2003-02-07 | ||
JP2003-058626 | 2003-03-05 | ||
JP2003058626 | 2003-03-05 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2004071130A1 true WO2004071130A1 (ja) | 2004-08-19 |
Family
ID=32852686
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2004/001261 WO2004071130A1 (ja) | 2003-02-07 | 2004-02-06 | 収音方法及び収音装置 |
Country Status (5)
Country | Link |
---|---|
US (1) | US7716044B2 (ja) |
EP (1) | EP1592282B1 (ja) |
JP (1) | JP4104626B2 (ja) |
DE (1) | DE602004006967T2 (ja) |
WO (1) | WO2004071130A1 (ja) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005136709A (ja) * | 2003-10-30 | 2005-05-26 | Nippon Telegr & Teleph Corp <Ntt> | 収音方法、装置、プログラム、および記録媒体 |
EP1647972A3 (de) * | 2004-10-08 | 2006-07-12 | Micronas GmbH | Verbesserung der Verständlichkeit von Sprache enthaltenden Audiosignalen |
WO2011004503A1 (ja) * | 2009-07-08 | 2011-01-13 | 株式会社日立製作所 | 雑音除去装置及び雑音除去方法 |
JP2011523836A (ja) * | 2008-06-02 | 2011-08-18 | クゥアルコム・インコーポレイテッド | マルチチャネル信号のバランスをとるためのシステム、方法及び装置 |
WO2012023268A1 (ja) * | 2010-08-16 | 2012-02-23 | 日本電気株式会社 | 多マイクロホン話者分類装置、方法およびプログラム |
JP4957810B2 (ja) * | 2008-02-20 | 2012-06-20 | 富士通株式会社 | 音処理装置、音処理方法及び音処理プログラム |
US8898056B2 (en) | 2006-03-01 | 2014-11-25 | Qualcomm Incorporated | System and method for generating a separated signal by reordering frequency components |
JP2017108441A (ja) * | 2012-11-12 | 2017-06-15 | ヤマハ株式会社 | ホスト装置およびマイクユニット |
Families Citing this family (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4701944B2 (ja) * | 2005-09-14 | 2011-06-15 | ヤマハ株式会社 | 音場制御機器 |
US20070083365A1 (en) | 2005-10-06 | 2007-04-12 | Dts, Inc. | Neural network classifier for separating audio sources from a monophonic audio signal |
EP2246846B1 (de) * | 2009-04-29 | 2018-06-06 | Deutsche Telekom AG | Schattenfilter-System für quasi-unterbestimmte blinde adaptive MIMO-Filterung |
EP2537350A4 (en) * | 2010-02-17 | 2016-07-13 | Nokia Technologies Oy | PROCESSING AN AUDIO RECORDING OF MULTIPLE DEVICES |
KR20120015218A (ko) * | 2010-08-11 | 2012-02-21 | 엘지이노텍 주식회사 | 데이터 재생 장치 |
KR101702561B1 (ko) * | 2010-08-30 | 2017-02-03 | 삼성전자 주식회사 | 음원출력장치 및 이를 제어하는 방법 |
GB2493327B (en) | 2011-07-05 | 2018-06-06 | Skype | Processing audio signals |
GB2495128B (en) * | 2011-09-30 | 2018-04-04 | Skype | Processing signals |
GB2495129B (en) | 2011-09-30 | 2017-07-19 | Skype | Processing signals |
GB2495278A (en) | 2011-09-30 | 2013-04-10 | Skype | Processing received signals from a range of receiving angles to reduce interference |
GB2495130B (en) | 2011-09-30 | 2018-10-24 | Skype | Processing audio signals |
GB2495131A (en) | 2011-09-30 | 2013-04-03 | Skype | A mobile device includes a received-signal beamformer that adapts to motion of the mobile device |
GB2495472B (en) | 2011-09-30 | 2019-07-03 | Skype | Processing audio signals |
GB2496660B (en) | 2011-11-18 | 2014-06-04 | Skype | Processing audio signals |
GB201120392D0 (en) | 2011-11-25 | 2012-01-11 | Skype Ltd | Processing signals |
GB2497343B (en) | 2011-12-08 | 2014-11-26 | Skype | Processing audio signals |
US9307335B2 (en) * | 2012-07-31 | 2016-04-05 | Japan Science And Technology Agency | Device for estimating placement of physical objects |
US9258661B2 (en) | 2013-05-16 | 2016-02-09 | Qualcomm Incorporated | Automated gain matching for multiple microphones |
JP6261043B2 (ja) * | 2013-08-30 | 2018-01-17 | 本田技研工業株式会社 | 音声処理装置、音声処理方法、及び音声処理プログラム |
KR101581619B1 (ko) | 2015-02-04 | 2015-12-30 | 서울대학교산학협력단 | 소리 수집 단말, 소리 제공 단말, 소리 데이터 처리 서버 및 이들을 이용한 소리 데이터 처리 시스템 |
JP6693340B2 (ja) * | 2016-08-30 | 2020-05-13 | 富士通株式会社 | 音声処理プログラム、音声処理装置、及び音声処理方法 |
CN107135443B (zh) * | 2017-03-29 | 2020-06-23 | 联想(北京)有限公司 | 一种信号处理方法及电子设备 |
KR102477099B1 (ko) | 2018-08-21 | 2022-12-13 | 삼성전자주식회사 | 소리 방향 탐지 센서 및 이를 포함하는 전자 장치 |
CN114566137A (zh) * | 2021-12-31 | 2022-05-31 | 苏州茹声电子有限公司 | 一种基于主动降噪的车辆路噪控制方法及***、存储介质 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH10190848A (ja) * | 1996-11-12 | 1998-07-21 | Lucent Technol Inc | 音響エコーキャンセル方法とそのシステム |
JPH1118194A (ja) * | 1997-06-26 | 1999-01-22 | Fujitsu Ltd | マイクロホンアレイ装置 |
JP2002149190A (ja) * | 2000-11-01 | 2002-05-24 | Internatl Business Mach Corp <Ibm> | 観測データから原信号を復元する信号分離方法、信号処理装置、モバイル端末装置、および記憶媒体 |
JP2002335194A (ja) * | 2001-05-09 | 2002-11-22 | Yamaha Corp | 2チャンネルエコーキャンセル用フィルタのインパルス応答設定方法および2チャンネルエコーキャンセラ並びに双方向2チャンネル音声伝送装置 |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2602750B2 (ja) | 1992-03-02 | 1997-04-23 | 日本電信電話株式会社 | 反響消去装置 |
JPH08250944A (ja) | 1995-03-13 | 1996-09-27 | Nippon Telegr & Teleph Corp <Ntt> | 自動音量制御方法およびこの方法を実施する装置 |
US6594367B1 (en) | 1999-10-25 | 2003-07-15 | Andrea Electronics Corporation | Super directional beamforming design and implementation |
JP3514714B2 (ja) | 2000-08-21 | 2004-03-31 | 日本電信電話株式会社 | 収音方法およびその装置 |
US7158933B2 (en) * | 2001-05-11 | 2007-01-02 | Siemens Corporate Research, Inc. | Multi-channel speech enhancement system and method based on psychoacoustic masking effects |
KR20030050338A (ko) * | 2001-12-18 | 2003-06-25 | 엘지전자 주식회사 | 가중치 갱신 방법 |
US7218741B2 (en) * | 2002-06-05 | 2007-05-15 | Siemens Medical Solutions Usa, Inc | System and method for adaptive multi-sensor arrays |
-
2004
- 2004-02-06 US US10/510,955 patent/US7716044B2/en active Active
- 2004-02-06 JP JP2005504891A patent/JP4104626B2/ja not_active Expired - Lifetime
- 2004-02-06 EP EP04708849A patent/EP1592282B1/en not_active Expired - Lifetime
- 2004-02-06 WO PCT/JP2004/001261 patent/WO2004071130A1/ja active IP Right Grant
- 2004-02-06 DE DE602004006967T patent/DE602004006967T2/de not_active Expired - Lifetime
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH10190848A (ja) * | 1996-11-12 | 1998-07-21 | Lucent Technol Inc | 音響エコーキャンセル方法とそのシステム |
JPH1118194A (ja) * | 1997-06-26 | 1999-01-22 | Fujitsu Ltd | マイクロホンアレイ装置 |
JP2002149190A (ja) * | 2000-11-01 | 2002-05-24 | Internatl Business Mach Corp <Ibm> | 観測データから原信号を復元する信号分離方法、信号処理装置、モバイル端末装置、および記憶媒体 |
JP2002335194A (ja) * | 2001-05-09 | 2002-11-22 | Yamaha Corp | 2チャンネルエコーキャンセル用フィルタのインパルス応答設定方法および2チャンネルエコーキャンセラ並びに双方向2チャンネル音声伝送装置 |
Non-Patent Citations (1)
Title |
---|
See also references of EP1592282A4 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005136709A (ja) * | 2003-10-30 | 2005-05-26 | Nippon Telegr & Teleph Corp <Ntt> | 収音方法、装置、プログラム、および記録媒体 |
EP1647972A3 (de) * | 2004-10-08 | 2006-07-12 | Micronas GmbH | Verbesserung der Verständlichkeit von Sprache enthaltenden Audiosignalen |
US8005672B2 (en) | 2004-10-08 | 2011-08-23 | Trident Microsystems (Far East) Ltd. | Circuit arrangement and method for detecting and improving a speech component in an audio signal |
US8898056B2 (en) | 2006-03-01 | 2014-11-25 | Qualcomm Incorporated | System and method for generating a separated signal by reordering frequency components |
JP4957810B2 (ja) * | 2008-02-20 | 2012-06-20 | 富士通株式会社 | 音処理装置、音処理方法及び音処理プログラム |
JP2011523836A (ja) * | 2008-06-02 | 2011-08-18 | クゥアルコム・インコーポレイテッド | マルチチャネル信号のバランスをとるためのシステム、方法及び装置 |
WO2011004503A1 (ja) * | 2009-07-08 | 2011-01-13 | 株式会社日立製作所 | 雑音除去装置及び雑音除去方法 |
JP5382745B2 (ja) * | 2009-07-08 | 2014-01-08 | 株式会社日立製作所 | 雑音除去装置及び雑音除去方法 |
WO2012023268A1 (ja) * | 2010-08-16 | 2012-02-23 | 日本電気株式会社 | 多マイクロホン話者分類装置、方法およびプログラム |
JP2017108441A (ja) * | 2012-11-12 | 2017-06-15 | ヤマハ株式会社 | ホスト装置およびマイクユニット |
US10250974B2 (en) | 2012-11-12 | 2019-04-02 | Yamaha Corporation | Signal processing system and signal processing method |
US11190872B2 (en) | 2012-11-12 | 2021-11-30 | Yamaha Corporation | Signal processing system and signal processing meihod |
Also Published As
Publication number | Publication date |
---|---|
EP1592282B1 (en) | 2007-06-13 |
EP1592282A4 (en) | 2006-08-09 |
EP1592282A1 (en) | 2005-11-02 |
DE602004006967T2 (de) | 2008-02-14 |
US20050216258A1 (en) | 2005-09-29 |
US7716044B2 (en) | 2010-05-11 |
JPWO2004071130A1 (ja) | 2006-06-01 |
JP4104626B2 (ja) | 2008-06-18 |
DE602004006967D1 (de) | 2007-07-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2004071130A1 (ja) | 収音方法及び収音装置 | |
US9210504B2 (en) | Processing audio signals | |
US8355510B2 (en) | Reduced latency low frequency equalization system | |
US9454956B2 (en) | Sound processing device | |
AU2004201374B2 (en) | Audio amplification apparatus | |
KR101934999B1 (ko) | 잡음을 제거하는 장치 및 이를 수행하는 방법 | |
US10319389B2 (en) | Automatic timbre control | |
JP2001309483A (ja) | 収音方法および収音装置 | |
JPH11298990A (ja) | オーディオ装置 | |
CN110111802B (zh) | 基于卡尔曼滤波的自适应去混响方法 | |
CN100534001C (zh) | 声音获取方法和声音获取装置 | |
US20130253923A1 (en) | Multichannel enhancement system for preserving spatial cues | |
JP2004078021A (ja) | 収音方法、収音装置、および収音プログラム | |
JP4116600B2 (ja) | 収音方法、収音装置、収音プログラム、およびこれを記録した記録媒体 | |
JP4568193B2 (ja) | 収音装置とその方法とそのプログラムとその記録媒体 | |
JP5801405B2 (ja) | 合成音声プロトタイプの推定 | |
JP2003250193A (ja) | 反響消去方法、この方法を実施する装置、プログラムおよびその記録媒体 | |
WO2018155164A1 (ja) | フィルタ生成装置、フィルタ生成方法、及びプログラム | |
JPWO2009025023A1 (ja) | 音像定位予測装置及び音像定位制御システム並びに音像定位予測方法及び音像定位制御方法 | |
JP3619461B2 (ja) | 多チャネル雑音抑圧装置、その方法、そのプログラム及びその記録媒体 | |
JP4298466B2 (ja) | 収音方法、装置、プログラム、および記録媒体 | |
US10135413B2 (en) | Automatic timbre control | |
CN117939360B (zh) | 一种用于蓝牙音箱的音频增益控制方法及*** | |
JP2014230085A (ja) | 音響処理装置 | |
CA2772322A1 (en) | Multichannel enhancement system for preserving spatial cues |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 2005504891 Country of ref document: JP |
|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): BW GH GM KE LS MW MZ SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2004708849 Country of ref document: EP Ref document number: 10510955 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 20048001742 Country of ref document: CN |
|
WWP | Wipo information: published in national office |
Ref document number: 2004708849 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 2004708849 Country of ref document: EP |